scholarly journals Mnemonic-Opto-Synaptic Transistor for In-Sensor Vision System

2021 ◽  
Author(s):  
Joon-Kyu Han ◽  
Young-Woo Chung ◽  
Jaeho Sim ◽  
Ji-Man Yu ◽  
Geon-Beom Lee ◽  
...  
Keyword(s):  
Gate Dielectrics ◽  
Vision System ◽  
Memory Function ◽  
Memory Cell ◽  
Sensor Nodes ◽  
Massive Data ◽  
Image Sensing ◽  
Sensor Processing ◽  
Artificial Neural Network Ann ◽  
A Charge

Abstract A mnemonic-opto-synaptic transistor (MOST) that has triple functions is demonstrated for an in-sensor vision system. It memorizes a photoresponsivity that corresponds to a synaptic weight as a memory cell, senses light as a photodetector, and performs weight updates as a synapse for machine vision with an artificial neural network (ANN). Herein the memory function added to a previous photodetecting device combined with a photodetector and a synapse provides a technical breakthrough for realizing in-sensor processing that is able to perform image sensing and signal processing in a sensor. A charge trap layer (CTL) was intercalated to gate dielectrics of a vertical pillar-shaped transistor for the memory function. Weight memorized in the CTL makes photoresponsivity tunable for real-time multiplication of the image with a memorized photoresponsivity matrix. Therefore, these multi-faceted features can allow in-sensor processing without external memory for the in-sensor vision system. In particular, the in-sensor vision system can enhance speed and energy efficiency compared to a conventional vision system due to the simultaneous preprocessing of massive data at sensor nodes prior to ANN nodes. Recognition of a simple pattern was demonstrated with full sets of the fabricated MOSTs. Furthermore, recognition of complex hand-written digits in the MNIST database was also demonstrated with software simulations.

scholarly journals A flexible ultrasensitive optoelectronic sensor array for neuromorphic vision systems

2021 ◽  
Vol 12 (1) ◽  
Author(s):  
Qian-Bing Zhu ◽  
Bo Li ◽  
Dan-Dan Yang ◽  
Chi Liu ◽  
Shun Feng ◽  
...  
Keyword(s):  
Sensor Array ◽  
Vision System ◽  
Vision Systems ◽  
Neuromorphic Circuits ◽  
Image Sensing ◽  
Perovskite Quantum Dots ◽  
Optoelectronic Sensor ◽  
Weak Light ◽  
Do So ◽  
Neuromorphic Vision

AbstractThe challenges of developing neuromorphic vision systems inspired by the human eye come not only from how to recreate the flexibility, sophistication, and adaptability of animal systems, but also how to do so with computational efficiency and elegance. Similar to biological systems, these neuromorphic circuits integrate functions of image sensing, memory and processing into the device, and process continuous analog brightness signal in real-time. High-integration, flexibility and ultra-sensitivity are essential for practical artificial vision systems that attempt to emulate biological processing. Here, we present a flexible optoelectronic sensor array of 1024 pixels using a combination of carbon nanotubes and perovskite quantum dots as active materials for an efficient neuromorphic vision system. The device has an extraordinary sensitivity to light with a responsivity of 5.1 × 107 A/W and a specific detectivity of 2 × 1016 Jones, and demonstrates neuromorphic reinforcement learning by training the sensor array with a weak light pulse of 1 μW/cm2.

Intelligent Vision Inspection System for IC Engine Head: An ANN Approach

2012 ◽  
Vol 479-481 ◽  
pp. 2242-2245 ◽  
Author(s):  
Rajesh Kanna ◽  
Manikandan Saravana
Keyword(s):  
Machine Vision ◽  
Vision System ◽  
Back Propagation ◽  
Engine Block ◽  
Inspection System ◽  
Back Propagation Algorithm ◽  
Ic Engine ◽  
Vision Inspection ◽  
System Module ◽  
Artificial Neural Network Ann

A machine vision system based on Artificial Neural Network (ANN) for inspection of IC Engine block was developed to identify the misalignment and improper diminishing of holes in the IC Engine block. The developed machine vision and ANN module is compared with the commercial MATLAB® software and found results were satisfactory. This work is broadly divided into four stages, namely Intelligent inspection module, Machine Vision module, ANN module and Expert system module. A system with a camera was used to capture the various segments of head of the IC Engine block. The captured bitmap format image of IC Engine block has to be filtered to remove the noises present while capturing and the size is also altered using SPIHT method to an acceptable size and will be given as input to ANN. Generalized ANN with Back-propagation algorithm was used to inspect the IC Engine block. ANN has to be trained to provide the inspected report.

An ANN Model to Estimate the Impact of Tea Process Parameters on Tea Quality

2015 ◽  
Vol 24 (09) ◽  
pp. 1550139
Author(s):  
Debashis Saikia ◽  
Diganta Kumar Sarma ◽  
P. K. Boruah ◽  
Utpal Sarma
Keyword(s):  
Process Condition ◽  
Correlation Study ◽  
Sensor Nodes ◽  
Quality Data ◽  
Inlet Temperature ◽  
Ann Model ◽  
Classification Rate ◽  
Tea Quality ◽  
Artificial Neural Network Ann ◽  
The Impact

Present study deals with the development of an artificial neural network (ANN)-based technique for tea quality quantification by monitoring fermentation and drying condition of the tea processing stages. An RS485 network-based instrumentation system has been developed and implemented for data collection for these two stages. Three calibrated sensor nodes are installed in the fermentation room due to its larger floor area to collect temperature and relative humidity (RH). Dryer inlet temperature is recorded using a calibrated thermocouple-based sensor node. From seven input parameters and target quality data obtained from tea taster, the ANN model has been developed to find the correlation between the process condition and the tea quality. From the correlation study, more than 90% classification rate is obtained from the model. The model is also validated with some independent data showing more than 60% correlation. Error in terms of root mean square error (RMSE) is about 0.17. This model will be helpful for improvement of tea quality.

scholarly journals Networking retinomorphic sensor with memristive crossbar for brain-inspired visual perception

2020 ◽  
Author(s):  
Shuang Wang ◽  
Chen-Yu Wang ◽  
Pengfei Wang ◽  
Cong Wang ◽  
Zhu-An Li ◽  
...  
Keyword(s):  
Visual Perception ◽  
Large Scale ◽  
Vision System ◽  
Image Sensor ◽  
Letter Recognition ◽  
Human Vision ◽  
Large Power ◽  
Image Sensing ◽  
Conventional Machine ◽  
Neuromorphic Vision

Abstract Compared to human vision, conventional machine vision composed of an image sensor and processor suffers from high latency and large power consumption due to physically separated image sensing and processing. A neuromorphic vision system with brain-inspired visual perception provides a promising solution to the problem. Here we propose and demonstrate a prototype neuromorphic vision system by networking a retinomorphic sensor with a memristive crossbar. We fabricate the retinomorphic sensor by using WSe2/h-BN/Al2O3 van der Waals heterostructures with gate-tunable photoresponses, to closely mimic the human retinal capabilities in simultaneously sensing and processing images. We then network the sensor with a large-scale Pt/Ta/HfO2/Ta one-transistor-one-resistor (1T1R) memristive crossbar, which plays a similar role to the visual cortex in the human brain. The realized neuromorphic vision system allows for fast letter recognition and object tracking, indicating the capabilities of image sensing, processing and recognition in the full analog regime. Our work suggests that such a neuromorphic vision system may open up unprecedented opportunities in future visual perception applications.

scholarly journals Machine vision based in-process light-emitting diode chip mounting system

2018 ◽  
Vol 51 (7-8) ◽  
pp. 293-303 ◽  
Author(s):  
Chao-Ching Ho ◽  
You-Min Chen ◽  
Po-Chieh Li
Keyword(s):  
Machine Vision ◽  
Printed Circuit Board ◽  
Vision System ◽  
Light Emitting Diode ◽  
Circuit Board ◽  
Robot Arm ◽  
Light Emitting ◽  
Printed Circuit ◽  
Delta Robot ◽  
A Charge

Background: In this study, a machine vision–based method was developed for automated in-process light-emitting diode chip mounting lines with position uncertainty. In order to place the tiny light-emitting diode chips on the pattern of a printed circuit board, a highly accurate mounting process is achieved with online feedback of the visual assistance. Methods: The system consists of a charge-coupled device camera, a six-axis robot arm, and a delta robot. The lighting system is a critical point for the in-process machine vision problem. Hence, designing the optimal lighting solution is one of the most difficult parts of a machine vision system, and several lighting techniques and experiments are examined in this study. In order to commence the mounting process, the light-emitting diode chip targets inside the camera field were identified and used to guide the delta robot to the grabbing zone based on the calibrated homography transformation. Efforts have been focused on the field of machine vision–based feature extraction of the chip pins and the holes on the printed circuit board. The correspondence of each other is determined by the position of the chip pins and the printed circuit board circuit pattern. The image acquisition is achieved directly online in real time. The image analysis algorithm must be sufficiently fast to follow the production rate. In order to compensate for the uncertainty of the light-emitting diode chip mounting process, a visual feedback strategy in conjunction with an uncertainty compensation strategy is employed. Results: Finally, the light-emitting diode chip was automatically grabbed and accurately placed at the desired positions. Conclusion: On-line and off-line experiments were conducted to investigate the performance of the vision system with respect to detecting and mounting light-emitting diode chips.

scholarly journals Mitigating Gray Hole attack in Mobile AD HOC Network using Artificial Intelligence Mechanism

2019 ◽  
Vol 8 (9S) ◽  
pp. 640-645
Keyword(s):  
Routing Protocol ◽  
Ad Hoc Network ◽  
Ad Hoc ◽  
Mobile Ad Hoc Network ◽  
Sensor Nodes ◽  
Destination Node ◽  
Radio Link ◽  
Mobile Nodes ◽  
Mobile Ad Hoc ◽  
Artificial Neural Network Ann

A mobile ad hoc network (MANET) is a combination of multiple mobile nodes, which are interconnected by radio link. In MANET, sensor nodes are free to move, and each node can act as a host or router. Routing is one of the most challenging tasks because nodes move frequently. Therefore, in MANET, the routing protocol plays an important role in selecting the best route to efficiently transmit data from the source node to the destination node. In this paper, the best path with efficient Ad Hoc on Demand Distance Vector (AODV) routing protocol is chosen as the routing mechanism. The properties of each node are categorized using firefly algorithm. The Artificial Neural Network (ANN) is trained as per these properties and hence in case if the gray hole node is detected within the route, it is identified and the route between the source and the destination is changed. At last, to show how effectively the proposed AODV with Firefly and ANN works is computed in terms of performance parameters. The throughput and PDR is increased by 4.13 % and 3.15 % compared to the network which is affected by gray hole attack. The energy up to 44.02 % has been saved.

scholarly journals Computer vision technique to classify dates based on hardness

2018 ◽  
Vol 22 (1) ◽  
pp. 36 ◽  
Author(s):  
Annamalai Manickavasagan ◽  
Naeema H. Al-Shekaili ◽  
Nawal K. Al-Mezeini ◽  
M. Shafiur Rahman ◽  
Negib Guizani
Keyword(s):  
Computer Vision ◽  
Vision System ◽  
Texture Features ◽  
Image Texture ◽  
Class Model ◽  
Linear Discriminant ◽  
Dried Fruits ◽  
Artificial Neural Network Ann ◽  
Vision Technique ◽  
Computer Vision Technique

Hardness is one of the important attributes in determining the quality of dried fruits. Hardness assessment is normally carried out by manual inspection. This method is time consuming, laborious, expensive and subjective. The objective of this study was to develop a computer vision system with a monochrome camera to classify dates based on hardness. Date samples were obtained from three different growing regions in Oman and graded into soft, semi-hard, and hard classes based on hardness. A total of 1800 date samples were imaged individually using a monochrome camera (600 dates / class). Histogram and texture features were extracted from the acquired monochrome images and used in the classification models. The overall classification accuracies in three class model (soft, semi-hard, and hard) were 66% and 71% for linear discriminant analysis (LDA) and artificial neural network (ANN), respectively. It was improved to 84% and 77% in LDA and ANN, respectively while using two class model (soft and hard (semi-hard and hard together)). The histogram features were more contributing in the date classification based on hardness than image texture features. Computer vision technique has great potential to develop online quality monitoring systems for dates and other dried fruits.

Hardware/Software Co-Design of a Vision System for Automatic Classification of Date Fruits

2020 ◽  
Vol 11 (4) ◽  
pp. 21-40
Author(s):  
Lazhar Khriji ◽  
Ahmed Chiheb Ammari ◽  
Medhat Awadalla
Keyword(s):  
Nearest Neighbor ◽  
Vision System ◽  
Texture Feature ◽  
Automatic Classification ◽  
Support Vector ◽  
K Nearest Neighbor ◽  
Artificial Neural Network Ann ◽  
Ann Classifier ◽  
Accuracy Performance

This paper proposes a hardware/software (HW/SW) co-design of an automatic classification system of Khalas, Khunaizi, Fardh, Qash, Naghal, and Maan dates fruit varieties in Oman. Three artificial intelligence (AI) techniques are used for qualitative comparisons: artificial neural network (ANN), support vector machine (SVM), and K-nearest neighbor (KNN). The accuracy performance of all AI classifiers is characterized for multiple color, shape, size, and texture feature combinations and for different critical parameter settings of the classifiers. In total, 600 date samples (100 dates/variety) are selected and imaged each sample individually. The system starts with preprocessing and segmentation of the colored input images. A total of 19 features are extracted from each image for use in classification models. The ANN classifier is shown to outperform all other classifiers. 97.26% highest classification accuracy is achieved using a combination of 15 color and shape-size features.

A Visually Servoed Active Camera for Miniature Mobile Robots

2000 ◽  
Author(s):  
Kemal B. Yesin ◽  
Bradley J. Nelson
Keyword(s):  
Motion Detection ◽  
Dynamic Capabilities ◽  
Vision System ◽  
Active Vision ◽  
Computational Power ◽  
Single Chip ◽  
Image Sensing ◽  
Detection And Tracking ◽  
Video Sensor ◽  
Active Camera

Abstract In this paper, an active vision system for a miniature mobile reconnaissance robot is presented. The system consists of a single-chip CMOS video sensor, a wireless video transmitter and miniature brushless D.C. motors. Visual tracking and servoing techniques were used to test the dynamic capabilities of the system. Additionally, a simple yet effective motion detection and tracking algorithm suitable for systems with limited computational power was developed and implemented. Available technologies for image sensing and actuation are surveyed for compatibility with the severe size, weight and power restrictions that the robot presents. The video system is designed to be concealed inside the robot and performs a deploy-retract motion in addition to pan and tilt. Different mechanism designs to reduce the number of actuators are presented.

Sign in / Sign up

Export Citation Format

Share Document